Temperature Measurements

General Questions and Comments that fit no specific category.
Post Reply
BillK

Temperature Measurements

Post by BillK »

My brew temperatures do not appear to match the temperature setting.

After I had my machine a few weeks, I spent some time with a Styrofoam cup and thermometer, and it made me suspicious that there was a difference between the set and indicated temperatures. So, I got out my thermocouple and Omega meter and did a number of more accurate measurements.

A brief description of what I did: I drilled a small hole in the middle of the single basket and threaded the thermocouple up into a location that is just about where the top of the coffee puck would be. I used orange silicone caulk to fasten it in place, and also to cover all of the little holes in the basket. I then made a calibrated hole that would let water out at the rate of 2 oz. in 30 seconds. I calibrated the thermcouple in a boiling water bath before I installed it. I used my bottomless portafilter to hold the basket.

I have taken quite a number of measurements right at about 200 deg. F - after the machine has sat idle overnight, after 15 minutes of inactivity, after an hour of inactivity, etc. I have found that I need to set the brew temperature at 95+3 in order to get a measured temperature of 199-201. So, as you can see, my measured temperature runs much lower than the setting.

I have my volumetric dosing set to 2 oz. This becomes 4 oz. if I give it a shot without any resistance. In order to heat up the brewhead prior to taking a temperature measurement I give it two shots without resistance, separated by the amount of time it takes the brew boiler to recover. I then take the measurement with the single basket apparatus in place. I take three measurements in a row, separated by the amount of time it takes for the brew boiler to recover. There is excellent correlation from measured shot to measured shot, never varying more than about 2 deg. F. Within a single measurement there is a ramp up to the maximum temperature, with the maximum temperature typically occurring late in the shot, as I would expect due to the water having to warm up the basket.

I have not run a full spectrum of measurements vs. settings, but I have done enough to know that the brew head does respond to different settings generally as would be expected, and I assume that there will be no problem other than that I will have to create a chart that translates each temperature setting to the measured value.

Does anybody see a problem with my setup? If so, I would like to get the setup corrected before I do more work.
User avatar
chas
Vivaldi Dreamer
Posts: 3044
Joined: Wed Jun 02, 2004 11:52 pm
Location: Central Maryland
Contact:

Post by chas »

Your test setup and methodology look fine to me. Have you checked out the measurements I took: http://www.rimpo.org/s1/S1Pix/S1TempMea ... NewBrd.pdf. Also Wolfgang Gaggl posted some he made here on the S1 Forum a couple of months back. You should be able to search and find them.

I also saw a first shot reading of 3C lower than the setting. A 2nd shot pulled as soon as boiler light went off was usually less than 1C lower and the 3rd and subsequent shots were pretty much dead on. The first shot was so far off that I didn't even record it in my data as I noted in the attachment.
Chas
LM GS/3 & LaSpaziale Dream v 1.25 (US 120V)
Mazzer Kony E, Customized Rocky
Hottop P/B
BillK

Post by BillK »

Yes, I have gone through your work and Wolfgang's. In both of your cases, you get correlation between your measured and set temperatures. I am just not getting it.

One thing I need to try is to put the portafilter and basket into boiling water just before I take a measurement so there is no cooling affect.
User avatar
chas
Vivaldi Dreamer
Posts: 3044
Joined: Wed Jun 02, 2004 11:52 pm
Location: Central Maryland
Contact:

Post by chas »

In doing your measurements you (and Wolfgang) are already emulating reality more closely that I did. Other than sacrificing the single basket by drilling a hole in it to put the TC wire through, all I did was to cut out a cylindrical piece of sponge to emulate the coffee puck and stick the TC through a hole in the middle. In this way the TC probe was about at the top dead center of where the puck would be and I could ensure that the hot water splashing off the top of the sponge would immerse the probe. I didn't do anything to restrict the flow rate to 2oz in 25 seconds.

I was surprised to get the consistency I saw at least after a shot or two. My goal when I put together that spreadsheet, was to use it when I wanted a specific temperature. I planned to look up what temp I needed to set the S1 at to get there.

In reality, except when I get the occasional coffee that has a very tight tolerance for temp=good coffee, I find that I like 92C and I am usually the only coffee drinker in the house so I just set it to 95C and use the first pull without doing any temperature flush shots first.

I - and I'm guessing most others when they first got the S1 - futzed around with the temperature settings obsessively, but now I've slacked off. Since I also roast my own beans so that I always start with absolutely fresh coffee, I can be off from the best that the S1 is capable of and still have coffee 1000% better than Starbucks. So unless I start getting shots that fall close to Starbucks quality I no longer mess with the temp.

Fortunately, I was like you when I first started. Othewise the S1 site and this Forum probably wouldn't exist :wink:
Chas
LM GS/3 & LaSpaziale Dream v 1.25 (US 120V)
Mazzer Kony E, Customized Rocky
Hottop P/B
wgaggl

Post by wgaggl »

One thing I want to mention: You seem to use an empty basket for your measurements?
Even if you restrict the flow through the bottom of the basket to get 2ml/sec, the empty basket needs to be filled. And that's been done with 9ml/sec. With my setup, where I take measurements every second (please see earlier posts, I guess you already did), I saw differences with empty vs. coffee filled basket, which can be attributed to the high flow rate at the beginning. So all my measurements are done 'in vivo', that is with the proper coffee dose.
Without doing these fast measurements to measure the entire profile, you'll be measuring averages, and probably don't see the difference between empty and filled basket (alternative is the Scarce PF).
Also, the calibration of your thermocouple is a crucial part, I also suggest taking into account, that the boiling point is always below 100degC! (correct for sealevel, and barometric pressure).

Other than that, you might see some difference, as the La Spaz is measuring boiler temp, not brewing temp, so you can expect a correction value. I haven't had the chance to test several different S1's for that purpose, but I guess there's some variability accross machines; it should be reproducable though for repeated measures on your machine.

The first shot you measure will be low, as Chas pointed out.

That's all that comes to my mind right now.

Cheers,
Wolfgang
BillK

Post by BillK »

Hi Wolfgang,

1. Do I have this correct? You are saying that an empty, flow-restricted portafilter will see a lower temperature reading than one with a puck in it that is restricted to the same flow because the initial rush of water in the empty basket case will cool the brew boiler. I do not see this at all with my setup. What I see is a temperature that rises within (as I recall) 5 seconds to a near maximum, followed by a slightly rising profile. It is very close to an L-shaped profile. My flow is restricted to 2 oz in 30-32 seconds.

2. I am at 1000 feet. I think that I computed at one point that water boils at my altitiude at about 210.5. My thermocouple typically reads somewhere around 210, which is actually closer than I would have expected, considering meter tolerance and the limits of the thermocouple.

3. Do you have any idea at all of what the typical drop is, from boiler temperature to what comes out of the shower screen? I did not kow that the temperature settings are calibrated to boiler temp as opposed to temp at the shower screen. Depending on this, it could explain all of what I think I am seeing.

4. Can you give me some idea of what you are seeing with your method, in terms of actual data? For example, do you have a final value of temperature for your machine when it is set at 95+3?

P.S When I got the "relax, lean-back and light-up" note from Chas, I decided he was absolutely right, why go through all the frustration of measuring temps like I did with my hx machine, and I put my meter and stuff away and just cranked the machine to the point where my shots taste best, and I haven't touched it since. But it would be nice to actually understand what is happening with the machine.

Thanks,
Bill
wgaggl

Post by wgaggl »

Hi Bill,

re:

1) I used a double basket, so the differences empty vs. filled are certainly bigger than with a single basket. I didn't see a difference in the average temp, but a temp profile, that's initially rising quickly, then slightly falling around 15secs, then slightly rising again (this 'dip' is below 1F overall difference). That dip is bigger was bigger for me using an empty PF. But you'd only notice it with continuous measurements.

2) Your boiling point would be at 209.6F for nominal barometric pressure.
See http://www.virtualweberbullet.com/boilingpoint.html for a good calculator.
And when you use boiling water to calibrate: Turn the heat up just bejond the point were bubbles are forming, and keep moving the probe around (steam bubbles are above boiling point whereas the water around is slightly below).

3) I'll go back over my notes on that (recently changed to a new PC and didn't transfer all my data yet). BTW: I don't know how La Spaziale calibrates their machines, but the probe sits in the group boiler and thus measures boiler temp. As Bob Roseman suggested correctly, the correction factor from boiler temp to group temp might not be a constant accross conditions (flowrates and set temperature, ...).

4) I average from 5+ seconds till end of shot to avoid averaging in the initial rise time of the L-shape to get a good representation of the stable period without the transient period at the begin. Again, I need to look back over my notes and maybe measure the drop at different set points. But roughly: I have my machine set at 95-1degC usually (Intelligentsia BlackCat and Kid'O) and observe a drop of about 1.5degC with very little variabiliy. And would expect that this offset changes from one S1 to the other.

As Chas noted, in the end it's just the taste that decides on the setting. So I wouldnt worry about a different offset value. Important is, that the temp doesn't vary too much across the shot. My timecourse in message http://www.rimpo.org/wforum/viewtopic.p ... light=#791 should only demonstrate the temp stability across the shot and the small range of variability.

Understanding the absolute temperature gets interesting when you try to compare the taste of a certain coffee blend across espresso machines. Right now I'm only using the S1, so I didn't put much effort into obtaining representative offset values at different set points of my S1. I was only concerned about creating reproduceable conditions for my setup.

Wolfgang
wgaggl

Post by wgaggl »

Bill, I went back to my measurement notes and also did another measurement to be sure.

Things about my methods, and issues with absolute temp measurement:

Calibration with boiling water: It's important to obain an average over a period of time while having the probe submerged into boiling water and keeping the water in constant motion (otherwise there will be a temp gradient in your pot). I submerged the probe of my Fluke 54/II, then sampled in 1sec intervals and averaged accross a minute while whirling the water.
Next, the boiling point of water is very much dependent on your barometric pressure, even more than elevation (http://www.biggreenegg.com/boilingPoint.htm). You can obtain the pressure in your area from http://www.weather.com. To be more precise, I have my barometer, and calibrated it based on the average difference between my barometer and weather.com for each day over a couple of weeks, to take local weather changes into account (the average however should be close to the average of the local weather station). This seems too scientific at first, but you'll notice that depending on the barometric pressure you might get up to 1degC difference in boiling point, and 1degC could make the difference between sour and bitter for certain espresso blends (assuming your technique has little variability) on the S1.
Taking a measurement: I drilled a little hole in the middle of a double basket, just wide enough to get the lead of the thermocouple through. The tip of the probe was just high enough to be level with the tamped puck (one could see the silvery tip of the probe, but didn't feel it with the finger), which was about 2mm below the shower screen. I did that with my bottomless PF to make sure I didn't see any channeling. Tamping with that probe takes some practice to get even extraction, no channeling and be as quick as without the probe. I started logging at the same time as I started the shot, the shot time always was between 25 and 30 sec, and logged a temp measurement every second.
Calulating the temp and offset: To obtain an average value, I calculated the mean from 5seconds into the shot till the end of the shot. This way the "L" shape at the begin of the infusion didn't have an effect on the value. For the offset I subtracted the average from the temp set value of the S1.

The average offset value between setpoint and actual temp for my S1 with the above methods is 3.6+/-0.11degC for 5 measurements at 94 and 95 degC setpoint. Note the small variability of averages from different shots (+/-0.11degC). The average standard deviation of temperature within shots (from 5sec till the end) was approx. +/-0.18degC, which shows the S1's great temp stability.

Now, I really want to point out, that absolute offsets may vary between machines, mostly because they depend on the calibration of the internal temp probe on the brewing group. Just as an example the max steam boiler pressure on different S1's varies from 1.2 to 1.6 bar (http://www.rimpo.org/wforum/viewtopic.php?t=152), and this is set by a temperature probe (pV=nkT), so there are temp probe calibration differences between machines for the steam boiler, why not also for the group boiler?

And as I pointed out in my methods, even with a precise digital logging thermometer like the Fluke 54/II, your absolute temp error might me substantial, depending on how many factors you take into account. And of course you'll end up with different values based on your methods.
So I would not put too much weight into absolute vaules. More important is temp stability across the shot and repeatablilty to the next shot.

I use absolute temp values only as a starting point and finetune it from there by taste. Also, it's sometimes good for troubleshooting, once you established ranges and 'norms' for your equipment and technique with a certain espresso blend.

OK, now I'm going to shut up :roll: :D
Just needed to point out these things I came accross during my quest for absolute temp measures and offset calculations.

Happy brewing,
Wolfgang
BillK

Post by BillK »

Wolfgang, what can I say? Wow!!

This is an outstanding study. I am SO impressed with your work!!

You should take great pride in this temperature study, Wolfgang, and you should go public with it beyond the S1 Forum. Beyond the value of demonstrating a first-rate method, it creates for the S1 a level of documented temperature control/repeatibility that is completely unrecognized. I am referring here to the exceedingly low standard deviation, as opposed to absolute temperature measured versus selected, which is far more important to what is achieved in the cup.

I do have a couple of questions:

First, depending on your flush-up procedure, my data pretty much reflects your findings. What was your flush-up procedure?

Second, you previously mentioned an anomaly in the temperature profile curve that you though I could not observe with my technique, which you thought was one of obtaining an average temperature. Actually, I used an Omega datalogger with temperature readings at one second intervals. My temperature profile never showed a drop at any point along it, whereas you indicated a momentary drop. Do your latest readings confirm a drop?
wgaggl

Post by wgaggl »

Thanks for the praise, Bill! :D

Flushing: I flush for about 3 seconds with the unloaded PF in. Then I fill and tamp. Doesn't take more than 15 seconds, even with the temp probe. During this time the temp light starts flashing (if it didn't immediately) and is ready once the PF is locked in.
You really hit an interesting variable to control for! I might look into this sometime.

Yes, I still see this little drop around the middle of the shot. In case of a well filled basket vs. an underfilled or empty basket, the drop is neglectably small though (about 0.1degC) and it's virtually invisible in the graphs. What I found interesting, that the temp profile I showed in http://www.rimpo.org/wforum/viewtopic.p ... light=#791, the first one I did, is kind of atypical, compared to what I described to you and see in all my recent measurements, but I noted there, that I didn't control for anything back then, unfortunately...

Wolfgang
BillK

Post by BillK »

My flush-up procedure for real, actual shots is a bit different from your test procedure, based on what I learned when I did my temperature measurements. I have found that to get to the most repeatable shot-to-shot temperature after a long idle I want to flush-up twice. (I don't think that what I have found has any bearing on your testing, BTW.)

When I do an actual espresso shot, I dose into the basket separately from the portafilter, so the way I do it is to hit the double shot button with the portafilter (no basket) in place. When the boiler has recovered, I do it again - two flush-up operations, total. When the boiler has recovered I pull the shot. Each double shot flush-up operation uses 4 oz of water w/o any resistance from a puck, so I am using 8 oz of water before pulling a shot with a dead cold head. My measurements indicated that this is not completely necessary, but sometimes it works better. Let me explain:

Since I used an artificial puck resistance for my measurements (with 30 sec time for 2 oz water), I was able to take repeated measurements as rapidly as the brew boiler would cycle - no puck to knock out and re-load. For a long-idle, cold head, I would do the following:

Flush-up w/o any flow resistance (4 oz. water)
Wait for brew boiler light
Flush-up, as before (4 oz. water)
Wait for brew boiler light
Pull (simulated) shot
Empty residual water from portafilter, replace, wait for boiler light
Pull 2nd (simulated) shot
Repeat as desired

What I found is that, most of the time, a single flush-up warmed up the head enough so that repeated, simulated shots had completely repeatable brew temperatures. However, sometimes the first simulated shot was a tad cool (cannot remember by how much, exactly). I have no idea why it was not always one way or the other.

So, given that I don't care how much water I use (love that plumbed-in, rotary pump!!), I have just gotten into the habit of flushing-up twice while I am preparing everything else. I don't think twice is necessary, but it takes no extra time, since the flush-up time is overlapped with my other activities.

As I see it, LaSpaziale screwed-up in a small way when they designed this machine. There is a more-or-less constant temperature offset between the brew boiler and the head, as confirmed by both your measurements and mine. The operator doesn't care what temperature the brew boiler is set to. The operator cares what the temperature at the grouphead is at. In other words, the temperature settings on the panel should be designed from the point of view of the operator, not the designer. This is easy enough for us to compensate for by optimizing taste in the cup, but they could have done it better. Do you see it this way, or do you think their approach is right?

You need to publish your work on one of more widely viewed forums, Wolfgang. Your work is very good.
User avatar
chas
Vivaldi Dreamer
Posts: 3044
Joined: Wed Jun 02, 2004 11:52 pm
Location: Central Maryland
Contact:

Post by chas »

There are a lot of variables to making great coffee:

1) Quality of the coffee including roast and age.
2) Tamp pressure
3) The approx. 25 second golden rule
4) Water temp and stability
5) Water pressure at the puck

The above discussion - since I assume everyone spending the cash on an S1 is using good coffee - covers the first four topics.

We've had some discussions about water pressure, how to test it, and how to adjust in on the S1 but it has been quite a while. Have most of you newer users sprung for a portafilter pressure gauge? If not I urge you to do so. Otherwise you are missing the other key variable to getting the best out of your S1.

The water pressure from the rotary pump represents a percent boost above the incoming "street" water pressure. So the pressure at your machine will be different than when set at the factory if their street pressure is significanlty different that it is at your home. In addition, there are different theories as to the best pressure. Many ascribe to 8.2psi as the ideal pressure for a rotary machine. Other say as high as 9.0psi. Mine was set to about 9.2 when I received it.

It would be great for someone to make some qualitative measurements of coffee quality versus group pressure on an S1. I've thought about it, but never gotten around to trying anything scientific. At one time it did drop the pressure from 8.2 to 7.5 for a while and couldn't definitively tell the difference. (Why I did that is the subject of a different discussion!)

My measurements were done with a stock PF pressure gauge. There have been a number of discussions on alt.coffee over the years about the proper way to measure the pressure. This requires installing a Tee fitting between the bottom of the PF and the guage then adding a needle valve off the Tee. The needle valve is then opened just enough for 2 oz of water to drain in 25 seconds. Then you are measuring the real pressure a puck sees not the pressure in a completely closed system.

Any takers? Any comments? Now I have myself going, I realise that it's been a while since I got out my gauge and checked my pressure. It's not a bad thing to do a couple of times a year.
Chas
LM GS/3 & LaSpaziale Dream v 1.25 (US 120V)
Mazzer Kony E, Customized Rocky
Hottop P/B
wgaggl

Post by wgaggl »

Bill,
I didn't do any measurements about different flushing procedures.

One thing to keep in mind: with an 800W heating element you can heat 2.7ml/sec. The open flow rate of the S1 is about 9ml/sec, which means every flushing shot cools the boiler, and if performed long enough, this translates to cooling the group too. Have you ever measured the temp profile during repeated flushing?

Simple calculation: The 800W heating element is capable of heating water by about 80F at a flow rate of 9ml/sec. Room temp is about 70F, you will end up with 120ml (your 4oz flush) water at 150F. Mixed with the other 330ml in the 450ml brew boiler at about 200F you end up with 186F at the group head by the end of your flush. It might actually be somewhat higher, since all the metal in boiler and group adds to the thermal mass, but you see my point: extensive flushing reduces the group head temp; if you keep a tight protocol, you might still end up getting more consistent temp profiles than without, but nevertheless, you might be cooling down your group head. The first few seconds of your flush take the group head temp up, the last ones may actually drop it down.

About the question about the S1's temp offset: I don't think that ever matters to most users, since they set the optimum temperature by preference of espresso taste; so did I. Only thing that really matters to the user is consistency, and as I pointed out earlier, that's where the S1 is quite good. For portability of measures between machines, you always need to measure temp profiles in the PF basket under resistance individually.

Wolfgang
wgaggl

Post by wgaggl »

BillK wrote: What I see is a temperature that rises within (as I recall) 5 seconds to a near maximum, followed by a slightly rising profile.
Bill,
I just looked back over your messages in this thread. Now that you told me the details about your flushing technique: Without having seen your data to look for other factors, the "followed by a slightly rising profile" suggests to me that you indeed cooled your PF down by flushing. And when the flow rate is below the max. of the heating element (<2.7ml/sec) during the shot, the group is heated up again.

I only flush briefly, primarily to synchronize the state of the group boiler cycle to my pulls, to gain consistency, secondarily to pull up the temp of the PF and basket (this matters only if the machine hase been sitting idle for a while). Sometimes, if the boiler has just recently heated, the light doesn't even immediately come on when I flush, it does within the following few seconds while I'm tamping; other times it comes on immediately if the boiler temp is at the lower end of its hysteresis. Point is, the boiler always had just heated when I pull my shot, but I don't want to cool it down by flushing (after all, it's not a HX).

Your flushing technique might still add consistency for you, because "if you do what you did, you get what you got". This phrase is typically used in a negative context, but it's also the very key to consistency.

Wolfgang
BillK

Post by BillK »

Hi Wolfgang,

I don
wgaggl

Post by wgaggl »

Bill,
I tried the flushing method you proposed (just for a couple of measurements though).

The average temperature I get interestingly is 1degC higher (this might need further inverstigation).

However, the standard deviation within a shot (5sec to end) is 23% higher than what I get by not flushing that much. So across a shot I doubt that your flushing strategy adds any notable benefit for flattening the temp profile.

Even commercial double boiler machines are not designed to keep the temperature when drawing that much water that quickly (Schomer also notes that), so I believe flushing should be kept brief.

You are correct that my water mixing example is oversimplified. However, the boiler is not a thin tube where incoming water doesn't mix with flushed water, so some mixing will occur, and the fact, that the heating element is only able to continuously reheat 2.7ml/sec from room to brewing temp remains. Every amount exceeding that will initially be buffered by the thermal mass of boiler (means it cools down the boiler) and group (depending on the amount of water been flushed), but the heat capacity of the metal is about 10 times less than the one of water. IMO you're sending your boiler on a temperature rollercoaster. But again, as long as you are consistent in what you are doing, it's always much better then having no repeatable procedure.

It's definitely worth looking into. I'm just always sorry for pulling more shots than I drink, so I didn't look into many factors yet. I'm happy to exchange experiences! A good flushing technique should ideally prep your machine to pull very reproduceable shots no matter if you pull one at a time or a dozen.

Thanks for your comments! :)

Wolfgang
BillK

Post by BillK »

Hi Chas,

Water pressure:

I did not know about the official alt.coffee method for measuring water pressure. I'll bet there isn't much difference between the method they are discussing and doing the easier measurement of the static head.

My setup is a little different. We have two places, and I move the S1 seasonally between them (this S1 never sees winter, except for the holidays :-)). I have two sets of filters-on-a-board, along with the Chriscoffee pressure regulators. The supply pressures at the two houses are definitely different. When I move the machine, I set it up and make the brew pressure adjustment with the external pressure regulator (don't know if this is as good as doing it in the machine, but it should be fine). My gauge is one that bought at a hardware store and screwed into one of my portafilters. I don't know if it is accurate, but it is highly repeatable. I set my pressure to 125 psi, typically. I have varied it from time to time between about 120 and 135, which I think it about 8.3-9.3 at my altitude (1000 ft, both places). I have never noticed much of a difference in the flavor, but I have never had completely stable conditions. I am certainly willing to ding around with this, but I will probably have to have exactly the same gauge as everybody else if the comparisons are to have any meaning. I think a possible first step might be good for somebody to establish what the difference in pressure is between the alt.coffee method and static pressure - either that or just ignore the alt.coffee method.
BillK

Post by BillK »

I don
wgaggl

Post by wgaggl »

BillK wrote:..., which is not going to be affected much by the flushing procedure (or lack of it), as long as the operator is consistent.

But one does have to flush-up after a long idle in order to achieve a temperature that is going to be consistent for multiple shots, if this is the goal.
Bill, I totally agree on these points. I think the S1 has really great consistency, as long as the operator is consistent.

Also I do agree that in order to be consistent with long idle times, flushing needs to be done. How much and how long is an interesting question (eventhough I'm already pretty happy with my S1), especially since I just pull a few shots a day (unless I host a party).

Wolfgang
Post Reply

Return to “General Q&A”